Effects of Previous Achievement and Metacognition on Knowledge Networks in Physical Chemistry: A Path Analysis
This dissertation tests a path diagram of physical chemistry learning consisting of three variables: previous achievement, metacognition, and knowledge network structure. Knowledge networks (Dearholt & Schvaneveldt, 1990) are visualizations of conceptual knowledge that appear as a web of connected words related to a topic. Networks can be analyzed according to structural features that measure word pair relatedness (coherence, similarity, and closeness), word string relatedness (betweenness), and highly relevant words (eigenvector centrality). Structural features in novice and expert networks can be compared to approximate expertise. Metacognition (the analysis of knowledge) is measured as Metacognitive Activities Inventory (MCAI; Cooper & Sandi-Urena, 2008) percentage scores. Previous achievement is measured in this study as previous chemistry, physics, and math course grades. Previous achievement was hypothesized to predict metacognition because students who draw from different types of previous knowledge to build physical chemistry knowledge are more likely to be metacognitive. Metacognition is hypothesized to predict knowledge network structure because metacognition is shown to influence learning behaviors, which affect the acquisition of conceptual knowledge. Conceptual knowledge of physical chemistry is the basis of knowledge networks. Statistical relationships among these path diagram variables were tested in a pilot study and in the main study.The pilot study tested the utility of knowledge network structure metrics in identifying differences among physical chemistry experts and novices. Three instructors validated the network word list which consisted of 3 physical chemistry terms. Two instructors and twelve students rated the degree of relatedness for a series of word pairs. Regression equations used previous achievement grade averages as independent variable(s) and knowledge network structure metrics as separate dependent variables. Results suggest a significant relationship between previous achievement in chemistry, physics, and math courses and the network metric of coherence. These findings represent a connection between the demonstration of prior knowledge and the consistency of word pair relatedness ratings.The main study tested the individual relationships among the path diagram variables as well as the degree of overall data-model fit. Two instructors validated words lists pertaining to three consecutive physical chemistry course units: (1) quantum mechanics; (2) molecular symmetry and spectroscopy; and (3) electronic transitions and molecular structure. Six instructors and twenty-nine students rated the similarity of word pairs for a single course unit. The statistical relationships among previous achievement, metacognition, and knowledge network structure were tested with path analysis. The results do not provide evidence that previous achievement predicts metacognition or that metacognition predicts knowledge network structure. These results do not provide evidence that the demonstration of chemistry, physics, and math knowledge affects the analysis of knowledge in physical chemistry. In addition, there is no evidence that degree of knowledge analysis predicts conceptual understanding of physical chemistry concepts. Future studies with this physical chemistry learning model may benefit from modifications that include a larger sample, an alternative metacognition instrument, and variable rearrangement.
Stats
Viewed 130 timesDownloaded 52 times